I'm Eric Schmidt, and my pleasure to introduce our next presenting company, Tango Therapeutics. We're delighted to have with us, Barbara Weber, the company's CEO. We also have Daniella, the company's CFO, in the front row here. So Barbara, Daniella, thanks for spending some time with us today. And as we were just talking about, Barb, if there's a day of the year that I would love to have Tango here at my conference, it would be today, because we're just coming off, obviously, some very important data from Amgen in the PRMT5 inhibitor space, which I'm sure we'll talk about. But before we get into those details and that granularity, just give us the high-level elevator pitch, if you would, on Tango and what it is you guys are up to.
So the big picture is that we really started Tango to address tumor suppressor genes, a whole new space that, outside of PARP, has not been addressed in oncology. That is to say, almost all of the approved oncology drugs in the last 10 or 15 years are targeting activated oncogenes. All those same cancers have tumor suppressor genes, but you can't directly target them because they're missing or they're inactivated, so this is a way to get at some of those other cancer genes.
Okay, and let's start then with your lead program, probably where you're spending the bulk of your time these days, which is a PRMT5 inhibitor for MTAP deletions. Just talk to us about, you know, the biology of MTAP and what's happening on a molecular level with regard to this pathway.
Yeah, and the biology for PRMT5 is complicated, right? What makes it synthetic lethal is the presence of an MTAP deletion. And what MTAP does when MTAP is deleted, a metabolite called MTA accumulates in cancer cells that partially inhibits PRMT5. That allows the ability to make a PRMT5 inhibitor that takes advantage of that partial inhibition and the presence of MTA to fully inhibit PRMT5 in cancer cells, but not in normal cells. And that's really the basis of everything in any precision oncology drug, is trying to identify something that's different between a normal cell and a cancer cell and kill the cancer cell without harming the normal cells.
Okay. And if we're talking about MTAP deletions in oncology, what's the size of that opportunity? What histologies might feature MTAP deletions more frequently?
It's big, and it's very common. So it depends on the histology, but anywhere from, say, 40% or so of glioblastomas have MTAP deletions, 15%-20% of lung cancers. Overall, 10%-15% of all solid tumors will have an MTAP deletion.
Okay. And then moving on to some of the strategies directed at PRMT5 inhibition previously. Maybe just a quick recap of what went wrong with the first-generation compounds.
Going back to what I was saying a minute ago about one of the key things about a successful oncology drug is to find something that's different between a normal cell and a cancer cell. The original the first generation of PRMT5 inhibitors didn't do that. They inhibited PRMT5, which is an essential gene, meaning if you inhibit it, it'll kill any cell. They inhibited PRMT5 equally in normal cells and in cancer cells, so that there was essentially no therapeutic index, and it was not possible to hit the target hard enough in cancer cells to not cause severe toxicity in normal cells.
Okay, and at this point, we've got a number of next-generation PRMT5 inhibitors that are cooperative binders. Maybe just give us a quick lay of the land there, and then we'll get into your programs after that.
So in terms of mechanism, they all are the same. They're all called MTA-cooperative. They bind when MTA is bound to PRMT5. They lock PRMT5 into an inactivated state, and they have this differential effect on cancer cells that are MTAP deleted versus not. In terms of the first three, that's us. We have two PRMT5 inhibitors currently in the clinic. One is our first one, which is TNG908. That one is brain penetrant. The second one, TNG462, it went into the clinic about a year behind 908, is more potent and selective than 908, but is not brain penetrant, so we're looking at those two. And then our competition, Amgen, which you mentioned, update yesterday, they went into the clinic probably the earliest of all of them, and then what was Mirati, now BMS.
And then AstraZeneca has one that's coming behind, I guess.
Correct
... at the end of the pack. At this point, would you say PRMT5 inhibition is validated as a strategy in oncology?
To me, 100%, right?
[audio distortion]
I think that is the question that's out there for investors, but I think 100% it's a validated clinical target. I believe these molecules are gonna be good, important oncology drugs. You know, there's a bit of skirmishing about that right now, and I think part of that is because they look a bit different than what people are used to from targeting oncogenic drivers. The biology's different, and the landscape of the effect on patients is different as well.
Different how so?
I think overall, the response rate is generally lower than what people have gotten used to. I mean, think about EGFR inhibitors or ALK inhibitors, where you're seeing upwards of 70%-80% response rates now. But the durability, I think, is what will differentiate these, and there's some interesting points to think about that are present in the Amgen update from yesterday.
Okay, maybe we'll talk about that unless... it sounds like you want to hold off on durability until we get a little bit more into the Amgen data.
Sure.
But before we go there, on a tumor-by-tumor type, is there a reason that PRMT5- certain tumors or histologies would be more sensitive to PRMT5 based on biology or MTA levels endogenously or anything of that sort that, that you're learning about?
...There's not an obvious reason that's been explained, but I think that is true, that there are some tumor types that are more sensitive than others. Most tumor types are sensitive, a few are the most sensitive, and in my mind, a few are the least sensitive. The explanation for that sort of has yet to emerge.
So that'll be empirical, that we just understand-
It's empirical, exactly.
That certain are more or less sensitive.
Nothing common in those tumor types yet that we can better understand.
Nothing at all. You know, we and others did a lot of preclinical work looking at the genetic architecture, other mutations, anything like that. It's just not possible. But it is interesting that some of the tumor types. Well, I think maybe all of the tumor types that are particularly sensitive to PRMT5 inhibitors are tough tumors to kill, you know? Things like cholangiocarcinoma, mesothelioma-
[audio distortion]
Lung cancer, those things are, those are not easy, solid tumors.
Just sticking with the pathway, is there any one thing that you think we really need to know or you want to know to move your programs forward? Is there a biologic unknown that you think, you go, "Gee, if we just solved for that or knew that, I'd be able to be a better drug developer?
I think I could always be a better drug developer. You know, it's a humbling job, as you know. But I think that for me, sitting where I sit, I feel very convinced about the data that are already out there, and of course, what I know from our own data that hasn't yet been disclosed, which is an important target. I think this is gonna be an important target. I think it's gonna be developable and approvable in pancreatic cancer, and in lung cancer, and a number of other tumor types. It's at least our molecules have the potential for being, I think, good combination partners, in addition to being good single-agent drugs because they're so well-tolerated, and we're really actively moving forward on those plans as well.
In terms of combination partners, one potential combination is within the pathway, a MAT2A inhibitor, and of course, Ideaya is doing that with Amgen. Is... What's your latest thinking on that combination?
Yeah, that combination, I think we've probably talked more about that combination in the last five years than anything, and have not prioritized that as a clinical priority. The reason for that is, as you said, it's an intra-pathway combination. So the idea is that by adding a MAT2A inhibitor, you can further inhibit PRMT5-
Mm
... directly on PRMT5. I think our feeling, well, not just our feeling, the preclinical data strongly support the fact that if you can inhibit PRMT5 fully with a single agent, you can get to the same place. It's a little harder to do without a picture behind me. But when we do those experiments, in order to show the synergy, which is real between PRMT5 inhibitors and MAT2A inhibitors, we have to cut the dose about in half of our PRMT5 inhibitors. We give full dose, we get the same effect as the combination. So going back to drug development, it's just way easier to start with a singlet when you go into other combinations than a doublet, and then you're instantly into triplets.
Then, in terms of the biology, is there a combination with PRMT5 that you think makes the most sense?
In terms of the biology, less so. The MAT2A is probably the most biologically driven, but thinking about genetically driven, there's a couple of really interesting ones that we are actively working on. The CDK4/6 inhibitors are of particular interest because all and almost all MTAP-deleted tumors also have a CDKN2A deletion, and there's strong synergy or combination benefit, I should say, preclinically. So that combination could be quite active and at least applicable to all patients with an MTAP deletion.
Are they-
Sorry, go ahead.
Are they chromosomally close to each other?
They're right next to each other.
Okay, thank you.
In fact, the reason that it's true is that CDKN2A is the driver deletion. That's a tumor suppressor gene that's commonly lost. MTAP is lost as a passenger next to it. We can get into that, too.
Yeah, yeah.
Amgen ran into a little bit of trouble-
[audio distortion]
... screening for that, for that reason. The other thing that's really interesting for me is orthogonal combinations of other oncogenic drivers, so particularly the RAS inhibitors and in pancreatic cancer. That's another gonna be a big focus for us.
Okay. So let's unpack this Amgen data. It's by far, I think, the biggest data set we've seen from a PRMT5 inhibitor. It's hot off the presses at ESMO just yesterday. I'm sure you haven't had full time to digest everything there, and there was a lot of detailed information, especially in the footnotes. But what are your first maybe high-level thoughts on what they've shown us?
So my first high-level thought was, "Great, this is excellent," right? This is continuing to be convincing data on the validity of the target. Second is, I was really happy to see all those lung cancers and pancreatic cancers in the green part of the waterfall plot. That's been a big question out there, of whether: Is this gonna be just a niche drug in some of these rare tumor types like cholangio and meso, or is this gonna really work in pancreatic cancer and lung cancer? And to me, that, those questions were... You know, those answers are really starting to unfold, and there's certainly no reason or previous evidence that it wouldn't. So I think that those were really important. As you know, I focus very much on the durability, and there's a lot of important information on durability in the Amgen data also.
So I was really happy to see it. You know, what we wanted from Amgen was data that was good, but not too good. And we feel like that's what we got.
So let's talk about the not too good. I assume you're referring to tolerability, GI nausea, but what are your thoughts there?
I think they clearly have GI toxicity. That's their dose-limiting toxicity, and that is obviously problematic for patients, but it is biologically important also, because the fact that they're having to stop dosing because of GI toxicity and not getting to on-target toxicity tells us that they're not hitting the target as well as it could be hit.
Okay, so they went up to 1,200, actually, they went up 1,600 mg, and then had to dial back down to 1,200 mg and called that their MTD. What biomarker or other data did you see at that lower 1,200 mg dose that leads you to think you can do better, from an efficacy standpoint?
Maybe I could actually, in that standpoint, talk about our own data and our own dose escalation, and I'll use TNG462 as the example here, our more potent and selective MTAP inhibitor. We went up to 600 mg once a day, with very good safety and tolerability until 600, and then we hit anemia and thrombocytopenia. We felt that was what we were looking for, right? Clean profile up to on-target toxicity, dialed back then to 200 and 300 mg. Again, clean and sort of now parsing through what's the optimal dose there to manage the hematox. I think the biomarker, if you will, is actually the on-target bone marrow toxicity.
Okay. If I were Amgen, I'd say: Well, you just have on-target hematoxicity. But, what do you see in their biomarker data that suggests that, you know, maybe you'd want to be more potent at hitting PRMT5 or suppressing the pathway to where it is that they're not hitting?
I think it's hard to know based on the biomarkers. The SDMA may be what you're asking about. I just don't think SDMA is a good biomarker. It's really the only PD biomarker we have, but I don't think it's sensitive enough to get into that range where we're seeing how whether or not PRMT5 has been fully inhibited or not. I think you lose SDMA signal well before you fully inhibit the enzyme, which is the problem.
Is there any way you can tell that your own hematox is on target versus off target?
That's a good question. I have to say, I guess it's somewhat circumstantial, right? That we have full SDMA inhibition, and we know from preclinical work and mouse studies that that's what we expected, so. And I guess to your point, you're right, somebody could claim: "Well, you just have on-target tox." But in the words of one of my favorite clinical pharmacologists: "You know, at some point, you wanna hit on-target toxicity, because show me a drug that doesn't do anything, and I'll show you a drug that doesn't do anything.
No reason to think that GI nausea is going to be a class effect of any sort?
No, I don't think so.
Okay. With regard to the Amgen data, just again, maybe going down another level, we did see some activity in lung and pancreatic. We didn't see much in glioblastoma, or at least three patients that I saw in that database. Is that your view as well, or any thoughts on... I know you're looking at, go ahead, with 908 GBM?
Yeah. I mean, I think it's a little bit hard to say. It is only three patients, and I think the key piece of information there is whether or not they're actually getting drug into the CSF, and that kind of remains to be seen.
They've alternatively called it a brain penetrant or maybe non-brain penetrant compound historically, so.
I think based on the preclinical data, we would call it that, too. It's just a matter of-
Which would you call it,
I think Amgen's molecule, based on preclinical data, is brain penetrant.
... and should be called that. The question is the translation between the animal data and the humans. Are they really-
Mm
... getting enough in?
So in terms of GBM, it's still unclear whether it's the wrong histology or the wrong drug, and-
I believe that there's nothing magical about GBM. You know, you'll remember, right, that melanoma, lung cancer, and maybe now pancreatic cancer also used to all be viewed as impossible to treat tumors. Eventually, you know, you find the right drug, and they start to fall. The issue with GBM is compounded by finding the right drug and getting it into the CSF, and that's the next challenge for the PRMT5 inhibitors, because I believe if we can get enough drug into the CSF, they'll respond like other tumors.
One comment I think the discussant made, and perhaps it was just a throwaway comment, was that GBM is such an aggressive and rapidly progressing tumor, and that PRMT5 inhibition may be a slower onset mechanism that takes a little bit of time to develop, that, you know, it might be a difficult histology. Is that not your view, or?
That, I think, is true, and as you know, it's particularly compounded by the fact that you've got a fast-growing tumor with a lot of swelling inside an enclosed space. So what we would want to see is some hint of activity, and then move to the adjuvant setting, as opposed to relapsed/ refractory, where it may be difficult. But again, I still believe that eventually, that will be possible.
Okay. Amgen did show us a number of different biomarker-based endpoints. I guess they were doing some IHC testing. Did I see that PRMT5 in the data set? You mentioned SDMA as well, which we don't think is a particularly good but... I mean, where's the field or where are you heading with regard to what is the right, you know, kind of companion diagnostic/biomarker to use to develop these products in terms of the pathway and full coverage?
SDMA, being a PD biomarker, I think, as I said, I don't think is particularly helpful. It tells you you're on target, but it's not particularly quantitative. I think the other IHC that they reference in the slide deck has to do with the fact that they were allowing patients on study initially with CDKN2A deletions that didn't have confirmed MTAP deletions. They went back and reanalyzed a number of the tumors with an MTAP IHC to tell whether or not those patients were actually MTAP deleted or MTAP wild type. What they found was that about a third of the patients that had CDKN2A deletions were MTAP wild type. Those patients would not respond, right?
And so what they then went on to explain, again, in the footnotes, was that those patients that are MTAP wild type were inadvertently added to the trial or in the toxicity assessments, but they're not in the efficacy assessments, which I think is fair, right? That's right.
How is your trial conduct different?
We always had required, as did Mirati, that the patients have MTAP deletion testing. I would say, like all genetic testing, there is some inherent error rate, and it's particularly tricky when you're looking for homozygous deletions. So in our study as well, we've gone back and looked with MTAP IHC, and we do have a few wild-type patients as well.
Okay. And maybe you just wanna set expectations with regard to what we'll see and when from your two studies, 462 and 908, just to make sure everyone's on the same page.
We continue to plan to have a data release this year. We have a smaller data set than Amgen for a couple reasons. One is just comparing directly Amgen versus 462, our more potent and selective molecule. That trial on AMG 193 started about a year and a half before 462, so the numbers of patients are smaller. I would say that in addition to that, it's mostly our dose escalation.
You said about 30-some-odd patients, 30+.
30-some-odd patients, right?
Okay.
In addition to that, our escalation, our expansion cohort started only a few months ago, so I think the data maturity will be less. And finally, as I've said, as many people know, enrolling lung cancer patients to any trial right now is very competitive. So I don't think that we will have enough lung cancer patients to really compare head-to-head, but that's why I was happy to see them having some responses there.
And 908?
908 has more patients. It started about six months after the Amgen trial, so around 65 patients there, about 10 of which are at the dose that we are expanding. And there will be enough GBM patients to address the question of whether it's active in GBM. And then we will finally, Sorry, finally, we will also say: "What are our plans going forward? Which of the two are we moving forward with in non-CNS tumors, and are we moving forward with 908 in CNS tumors?
The only reason to move 908 forward would be in a CNS tumor indication?
That's always been our hypothesis, that one of the two, we would be able to select, 908 versus 462 for non-CNS indications. Now, if they were the same, then you might wanna use 908, right? Because it might be able to manage CNS metastases. If there was a clear advantage from an efficacy, durability standpoint, you might wanna go with 462.
Okay. And maybe in terms of the profile that we can expect to see, and I know there's, you know, gonna be limited amount of information you can share today. First, the data are coming out toward year-end, and I think you've also said in a company-sponsored forum. But with regard to GI and nausea, doesn't sound like that's been a concern. You already alluded to durability being really critical for this pathway, and I'm wondering first what you saw in the Amgen data set that encouraged you with regard to durability and how we might perceive that in your own data set.
I'd start by saying, if you look at the swimmer plot, which is just the dose escalation patients, there's a number of those patients that have crossed the one-year barrier, that have been on for you know, anywhere from 12 to 16, 18 months, which is pretty remarkable in and of itself. But the other bigger signal to me-